Personalized Health and Care: IT-enabled Personalized Healthcare


Published on

Healthcare reform currently focuses on changing the structure and incentives of the U.S. healthcare system. Healthcare transformation requires a more open, robust health information technology (HIT) environment to go beyond removing waste and inefficiencies to discover the science of health and care. Learn how IBM can make this possible.

Published in: Technology
1 Comment
No Downloads
Total views
On SlideShare
From Embeds
Number of Embeds
Embeds 0
No embeds

No notes for slide

Personalized Health and Care: IT-enabled Personalized Healthcare

  1. 1. IBM Global Business Services Healthcare Executive Report IBM Institute for Business Value IT-enabled personalized healthcare Improving the science of health promotion and care delivery
  2. 2. IBM Institute for Business Value IBM Global Business Services, through the IBM Institute for Business Value, develops fact-based strategic insights for senior executives around critical public and private sector issues. This executive report is based on an in-depth study by the Institute’s research team. It is part of an ongoing commitment by IBM Global Business Services to provide analysis and viewpoints that help companies realize business value. You may contact the authors or send an e-mail to for more information.
  3. 3. Introduction By Jim E. Adams, Edgar L. Mounib and Amnon Shabo To succeed in transforming healthcare, many countries will need to move to more personalized healthcare (PHC). Successful migration must encourage innovation, provide access to more complete patient information and incorporate advanced clinical knowledge into clinical decision making. Therefore, PHC will require a much more open, robust health information technology (HIT) environment than exists today. We have identified five major HIT-related challenges, as well as recommendations to foster HIT-enabled PHC. Executive summary PHC could help address difficulties associated with the science Healthcare systems around the world are making great strides of health promotion and care delivery by using broader and in technological, scientific and clinical innovations. Even so, deeper patient information and applying more complete many countries, even those with reputations for excellent care, clinical knowledge to help promote patient-centered health are struggling to address increasing costs, poor or inconsistent and predict, prevent, aid in early detection of, treat and manage quality and inaccessibility to timely care. Many believe that diseases. Through improved science, PHC has great potential fundamental transformation is required for what are becoming to improve quality and reduce overall costs of health increasingly unsustainable healthcare systems.1 promotion and care delivery. However, it is incredibly informa- tion and knowledge intensive even compared to today’s Three factors contribute to the unsustainability of healthcare: complex needs, which already exceed human cognitive capacity. fragmentation, waste and inadequate science for health Access to and appropriate use of burgeoning volumes of promotion and care delivery. Issues with fragmentation and patient information and clinical knowledge will require a waste are indeed daunting; they are a key focus of current U.S. powerful health information technology (HIT) environment. health reform efforts, for example. Receiving less attention is inadequate science – more explicitly, problems involving the A much more open, robust, flexible, standards-based HIT envi- science of health promotion and care delivery. These issues ronment will be required to enable personalized healthcare. impact both quality and costs, with estimates for unwarranted This environment must be capable of capturing, storing, care – just one part of inadequate science – ranging from analyzing and appropriately sharing information about US$250 to $325 billion per year in the United States.2 This individual patients and patient populations. It must be capable paper focuses on these inadequate science problems, which of rapidly generating new clinical knowledge, managing that present significant barriers to realizing the vision and promise knowledge and easily incorporating the knowledge into clinical of PHC. processes and workflows for decision making for health promotion and care delivery.
  4. 4. 2 IT-enabled personalized healthcare This environment also must facilitate appropriate interactions These challenges are much more difficult to address than the among constituents, whether they involve patients communi- HIT-related issues associated with healthcare’s fragmentation cating with their care delivery teams, communications among and waste problems. They are also more complex than the care providers or researchers working across traditional IT-related problems faced in other industries. Solutions will organizational, industry or country boundaries. However, these require sophisticated use of existing IT-related capabilities, as capabilities were not top priorities when today’s HIT systems well as the development of new approaches. were designed and implemented. Today’s systems were designed primarily to facilitate administrative functions such as While a robust HIT environment is necessary for PHC billing and payments and to automate specific clinical encoun- implementation, it is certainly not a panacea for success. Other ters such as a doctor’s appointment or hospital inpatient stay. hurdles, including those relating to policy, funding, education and ethics, must also be cleared. However, perhaps the first To realize the vision of PHC, five interdependent HIT-related step is ensuring stakeholders have a clear understanding of challenges must be overcome: PHC and its implications, followed by recognition that it must be a key part of the solution. 1. Lack of an interoperable HIT environment for care delivery and research 2. Prevalence of tightly coupled applications and data 3. Inadequate data and knowledge standards 4. Insufficient analytics capabilities “Estimates suggest that as much as US$700 5. Absence of a clinical decision-making foundation. billion a year in healthcare costs do not improve health outcomes. It occurs because we pay for more care rather than better care. We need to be moving towards a system in which doctors and hospitals have incentives to provide the care that makes you better, rather than the care that just results in more tests and more days in [the] hospital.” Peter Orszag, director of the White House Office of Management and Budget, during a National Public Radio interview 3
  5. 5. IBM Global Business Services 3 The need for PHC Many healthcare systems today are not really “systems” at all. Table of Contents They are fragmented, resulting in poorly coordinated patient The need for PHC 3 care and lack of accountability for overall costs and quality. It’s estimated this fragmentation costs the United States US$25 to Science of health promotion and care delivery 4 $50 billion annually.4 The accompanying lack of transparency Information and knowledge: Key for high-performance into costs and quality makes it difficult to be an informed healthcare system 6 healthcare services consumer. The U.S. system further exacer- PHC scope and vision 6 bates these problems through a reimbursement system that rewards volumes of procedures, particularly major acute Delivering PHC 8 interventions, instead of value. PHC with and without “-omics” 10 The value of IT-enabled PHC 12 Many healthcare systems also suffer from tremendous waste (spending that can be eliminated without reducing the quality of Exceeding human cognitive capacity 13 care) resulting from clinical and administrative inefficiencies.5 In The current state of HIT 14 the United States, clinical waste includes factors such as ineffi- Toward a new HIT environment 17 cient, error-prone, labor-intensive processes (costing US$75 to $100 billion per year); duplicate diagnostic testing due to Challenge 1: Lack of an interoperable HIT environment unavailability of results; and defensive medicine, coupled with for care delivery and research 18 high levels of fraud and abuse (estimated to be US$125 to $175 Challenge 2: Prevalence of tightly coupled applications billion per year).6 And administrative inefficiencies are and data 21 estimated to cost US$100 to $150 billion per year.7 Challenge 3: Inadequate data and knowledge standards 22 Challenge 4: Insufficient analytics capabilities 24 Again, these fragmentation and waste issues are at the heart of current U.S. health reform efforts (see Figure 1). However, this Challenge 5: Absence of a clinical decision-making paper’s focus is on problems involving the science of health foundation 26 promotion and delivery, which receive far less attention. To Summary: Key capabilities to address HIT challenges 30 achieve affordable, high-value healthcare that can appropri- Recommendations for stakeholders 30 ately tailor health promotion and care delivery to meet the needs of each individual – in other words, PHC – these Conclusion: The PHC journey 33 inadequate science problems must be addressed in large part. About the authors 34 Special content contributors 34 Acknowledgements 35 References 36
  6. 6. 4 IT-enabled personalized healthcare and environment. Approximately 80 percent of coronary Higher artery disease, up to 90 percent of type 2 diabetes, and 30 to • Administrative 70 percent of cancers could be prevented or significantly inefficiencies: System US$100 to $150 billion Science of delayed through lifestyle changes such as proper diet, structure • Clinical inefficiencies: health adequate exercise, limiting alcohol consumption and not and US$75 to $100 billion promotion incentives • Fraud and abuse: and care smoking.8 Even so, the United States is experiencing a Relative nontechnical difficulty US$125 to $175 billion delivery • Defensive medicine diabetes epidemic with 24 million diabetics and 57 million • Poorly coordinated care: • Activating lifestyle and prediabetics.9 behavior changes US$25 to $50 billion • Managing chronic diseases • What can be done to better manage the growing the number of • Volume-based rewards Waste and • Focusing on prevention, • No one is accountable inefficiency prediction/early detection, people with chronic diseases? Chronic disease accounts for about for value • Lack of transparency diagnosis and prognosis • Knowing when to treat with 75 percent of the costs of U.S. healthcare; yet, much of the early detection system remains oriented to providing acute care.10 Managing • Completely and accurately diagnosing chronic conditions requires a number of activities beyond • Knowing what works, even for populations acute care such as coordinating care appropriately among • Consistently applying what we know works clinicians; ensuring that the right preventive, diagnostic and Lower therapeutic interventions occur at the right time; monitoring Relative technical difficulty Higher and improving patient adherence to treatment regimens; Primary focus of current reform efforts activating lifestyle changes; and monitoring outcomes over Largely unaddressed by current reform efforts time. Sources: IBM Global Business Services; IBM Institute for Business Value. • What can be done to promote a greater focus on and effectiveness of Figure 1: HIT investments addressing structure and waste problems prevention and prediction/early detection both in individuals and can be beneficial but insufficient to address science problems. throughout the healthcare system? It is estimated that 56 percent of the chronically ill are not receiving appropriate preventative Science of health promotion and care delivery services.11 The problems associated with the science of health promotion • As early detection capabilities improve with tools such as advanced and care delivery involve a number of largely unanswered imaging, how does one learn more about which factors detected questions: require treatment or ongoing monitoring and which do not? For example, researchers are still learning which types of prostate • How can individuals be motivated to make better health and and other cancers need to be treated aggressively and which healthcare choices? Some diseases, such as cystic fibrosis or types can be monitored or treated less aggressively. Huntington’s disease, can be attributed directly to genetic • How can doctors ensure complete and accurate diagnoses? The rate variations and sometimes to a variation of a single gene. The of diagnostic error is up to 15 to 20 percent, and the cases great majority of disorders, however, such as cardiovascular physicians see as routine and unchallenging are often the ones disease, diabetes or cancer, are caused by complex interplay that end up being misdiagnosed.12 among multiple genes and nongenetic factors such as lifestyles
  7. 7. IBM Global Business Services 5 • What can be done to increase knowledge about which diagnostic and Percent of patient population for which drug in a class is therapeutic approaches work in real-world settings? It is estimated ineffective (on average) that only about 25 percent of care decisions are supported by Antidepressants (SSRIs) 38% (Cost: US$2.3 to $5.8 billion) evidence – and existing evidence-based knowledge tends to be fragmented and inaccessible.13 For chronic conditions, Asthma drugs 40% (Cost: US$.56 to $1.0 billion) prominent researchers estimate that evidence-based guidelines Diabetes drugs 43% exist for 20 to 33 percent of healthcare spending.14 Yet only about one-tenth of 1 percent of U.S. healthcare spending is Arthritis drugs 50% devoted to determining what works best.15 Alzheimer’s drugs 70% • What can be done to more consistently apply what is known to work? Despite having only limited clinical knowledge, patients in the Cancer drugs 75% United States receive only 50 percent of recommended Source: Spear, Brian B., Margo Heath-Chiozzi and Jeffrey Huff. “Clinical preventive, acute and long-term healthcare.16 Additionally, application of pharmacogenetics.” Clinical Trends in Molecular Medicine, Volume 7, Issue 5. May 1, 2001. even if perfect and complete clinical knowledge existed to address these questions about the science of health promotion Figure 2: Major drugs are ineffective for many. and care delivery, the current health information technology (HIT) environment could not help enable the consistent incorporation of this knowledge into clinical decision making. The fragmentation, waste and science problems contributing A simple example to illustrate the problems with the science of to the unsustainability of healthcare systems are related. If care delivery involves how different drugs affect different improvements are not made in the science of health promotion individuals. Major drugs are ineffective for many due to and care delivery, then other steps might be taken to lower differences in the way patients metabolize the drugs and the costs, which could negatively affect quality. For example, in an difficulty of identifying which disease might be causing a attempt to reduce waste or to address fragmentation by particular set of symptoms that are widely shared among coordinating care, decisions might continue to be made multiple diseases (see Figure 2). Drugs designed for Alzheim- without an understanding of what preventive, diagnostic or er’s disease, for instance, are effective for just 30 percent of the therapeutic interventions are the most effective or cost-effec- patient population – but doctors are unable to identify the 30 tive for patient populations or subpopulations. Innovation percent in advance.17 Prescribing the drugs to everyone could also be stifled if decisions are made to not pay for new, identified with Alzheimer’s is expensive and increases the risks perhaps initially more expensive, diagnostic or therapeutic for side effects. Better clinical knowledge would enable a capabilities due to a lack of understanding of the associated clinician to say to a patient, “This drug is 100 percent effective benefits or the total costs associated with not paying for the for 30 percent of the population and, based on our testing, we new capabilities (for example, ineffective treatments based on know that you are part of that 30 percent.” an inaccurate or incomplete diagnosis). In short, improving the science of health promotion and care delivery is critical to high-value health reform but has received insufficient attention thus far.
  8. 8. 6 IT-enabled personalized healthcare Information and knowledge: Key for high-performance expressed relating to different aspects, such as the govern- healthcare system ment’s ability to implement the provisions of the HITECH Better information and knowledge provided in large part Act, the criteria for physicians or hospitals to receive funding through HIT can help address all three sets of factors – frag- and the benefits that may ultimately accrue. mentation, waste and inadequate science – plaguing the U.S. and many other health systems. We believe improving the As governments and organizations invest in HIT, key questions science of health promotion and care delivery will require a arise, which this study seeks to answer: much more powerful HIT environment than one required to address the fragmentation and waste factors – but the clinical • What are the scope of and vision for PHC, and how can PHC cost and quality benefits could be considerable. Fortunately, improve the science of health promotion and care delivery? the HIT-related investments made to address fragmentation • How does PHC differ from today’s predominant health and waste factors can lay the foundation to continue to promotion and care delivery approaches? improve the science of health promotion and care delivery. For • How can PHC help address the cost and quality problems example, the electronic data generated by today’s electronic with today’s healthcare systems? health records (EHRs) can be used to some degree to generate • Are current HIT-related initiatives and investments sufficient knowledge required for better clinical decision making. to enable the continued migration to PHC? If not, what key challenges exist and how can they be addressed? Many countries are investing to improve their HIT environ- • What are the recommendations for key stakeholders? ments. In the United States, the Health Information Tech- • Will the HIT environment being funded by the U.S. nology for Economic and Clinical Health (HITECH) Act, part HITECH Act enable or inhibit progress toward PHC? of the American Recovery and Reinvestment Act (ARRA) of 2009, has allocated tens of billions of dollars toward enhancing PHC scope and vision the HIT environment, particularly implementing EHRs and In 2008, the U.S. President’s Council of Advisors on Science health information exchanges (HIEs) and supporting compara- and Technology (PCAST) defined personalized medicine (PM) tive effectiveness research to identify what works for which as “the tailoring of medical treatment to the individual patients under what circumstances.18 Even though almost all characteristics of each patient. It does not literally mean the agree that HIT investments are needed, concerns have been creation of drugs or medical devices that are unique to a patient, but rather the ability to classify individuals into subpopulations that differ in their susceptibility to a particular disease or their response to a specific treatment. Preventive or HIT-related investments addressing therapeutic interventions can then be concentrated on those fragmentation and waste can lay the who will benefit, sparing expense and side effects for those who will not.”19 foundation to continue to improve health promotion and care delivery.
  9. 9. IBM Global Business Services 7 For this paper, we will use this definition of personalized Third, PHC requires closely linked research and care delivery. medicine as a foundation for our broader term: personalized Clinical knowledge must continuously be generated, managed healthcare, or PHC. PHC expands on this PM definition in and appropriately incorporated into future decision making four key areas. First, PHC is broader than PM in scope. both for the individual and for similar patients and individuals. Although diagnosis and preventive or therapeutic interventions Also, clinical data must be appropriately and securely available for disease are critical in both PM and PHC, PHC also for research to advance the understanding of disease and appropriately emphasizes health promotion and ongoing treatment outcomes. monitoring and management of patients (see Figure 3). These additional areas become increasingly important with the Fourth, it must be participatory – individuals must be strong growing prevalence of chronic disease. participants in their own health and healthcare. The interplay of multiple factors affects an individual’s health, including Second, PHC is broader than medicine based on “-omics.” lifestyles and behaviors, unique human biology or genetic While PM is frequently and appropriately associated with makeup, the environment and medical care received. Individ- genomics (the study of genes and their function), proteomics uals make many health-related decisions outside the clinical (the study of the full set of proteins encoded by a genome), care setting, such as lifestyle choices or whether to adhere to epigenomics (the study of chemical compounds that modify, or treatment regimens. Additionally, individuals should participate mark, the genome in a way that tells it what to do, where to do appropriately in some clinical decisions through mechanisms it and when to do it) and other types of “-omics,” much can and such as advance directives or shared clinical decision making needs to be done, as we will discuss later, to personalize for conditions for which there is no one best treatment option. healthcare as knowledge of “-omics” expands.20 Integration of research and care delivery • Family history • Social history • Patient preferences • Genotypic information • Health risk assessment • Shared decision making • Other “-omics” information Risk Prevention Targeted Diagnosis Therapy Clinical assessment monitoring Discovery research/ Outcomes develop- Early detection testing and Manage and monitor new ment monitoring for new clinical patient information for clinical knowledge knowledge • Medical history • Patient-reported data/ • Environmental factors • Problems list questionnaires • Lifestyles/behaviors • Phenotypic information • Patient monitoring data • Test results • Compliance monitoring Clinical research Sources: IBM Global Business Services; IBM Institute for Business Value. Care delivery Figure 3: PHC tailors preventive, predictive, diagnostic and therapeutic activities to the specific characteristics of each patient.
  10. 10. 8 IT-enabled personalized healthcare In short, PHC helps address the problems with the science of More art than science health promotion and care delivery by using more individual- More science than art ized patient information and clinical knowledge to help Good Personalized promote patient-centered health and predict, prevent, aid in Access to relevant patient information (based on me or people like me) early detection of, treat and manage diseases. Evidence-based Delivering PHC (Based on patient populations) Very little health promotion and care delivery today would be Clinician considered PHC. Good decisions by patients and clinicians consensus-based about health promotion and healthcare depend heavily on two critical factors: access to relevant patient information and the ability to apply the best clinical knowledge. The first includes a g Intuitive medicine in as (Based on complete access to broad array of information, such as a patient’s family history, c re available patient information and In lue clinical knowledge) lifestyle, previous medical history, personal preferences and – va increasingly – individual genetic information. The second Trial and error (Based on clinician expertise and experience) involves knowledge that has been gained about how to Poor Access to clinical knowledge Good promote health, prevent disease, predict risk, diagnose (e.g., diagnostic tools, knowledge of causes of diseases, completely and correctly, and treat and manage patient empirical evidence or comparative effectiveness) conditions successfully. Sources: IBM Global Business Services; IBM Institute for Business Value. Unfortunately, medicine generally has been practiced with far Figure 4: The ability to personalize healthcare requires better access too limited availability of both patient information and clinical to a wider variety of relevant patient information and clinical knowledge. knowledge. It is helpful to visualize the evolving practice of medicine toward PHC as illustrated in Figure 4. The large diagonal arrow indicates a progression from practicing “one-size-fits-all” approach); and, finally, to PHC based on medicine based on the knowledge and experience of individual personal choices, where appropriate, and evidence developed practitioners; through intuitive or consensus-based approaches from smaller populations (also called subpopulations) as similar when evidence is sparse; to evidence-based approaches to the individual patient as possible. grounded in large populations (sometimes referred to as a Much of care delivery today is based on “trial and error” – the expertise and knowledge of the individual clinician, with limited access to relevant patient information and clinical knowledge that is not already “in the clinician’s head.” This PHC can help address problems associated approach, which is based more on art than on science, has led with the science of health promotion and care to many costly problems, including incorrect or incomplete diagnoses, use of ineffective interventions and failure to use delivery. effective interventions. This frequently is not a shortcoming of individual physicians. Rather, until recently, longitudinal patient information was not systematically captured and stored
  11. 11. IBM Global Business Services 9 in a way that was efficient or readily available to caregivers. disease, not just the symptoms.22 Additionally, PHC can help Further, insufficient tools exist to ensure clinicians have access tailor other health-related activities such as health promotion, to the latest clinical knowledge – and those that do exist may prevention or care management for the individual, thereby not contain the latest knowledge, given the pace and dynamism improving the outcomes. of today’s scientific discovery. When appropriately applied to an individual with a disease, Intuitive medicine differs from “trial and error” medicine in PHC can significantly improve the ability to assess risks, that the clinician has complete access to available patient understand events that initiate a disease and detect those events information and relevant clinical knowledge. Even so, molecularly (perhaps long before clinical detection is possible) knowledge about many diseases or combinations of diseases is and to tailor treatments specifically for that individual (see not sufficient to standardize the approach to diagnosis and Figure 5). This approach can improve the likelihood of treatment. Therefore, the clinician must depend largely upon preventing or reversing the disease, thereby reducing overall personal expertise, skills and intuition (insights and judgment, costs associated with the disease (see sidebar, Applying PHC to for example) to determine what is wrong (frequently based on breast cancer). symptoms) and how best to treat with therapies whose efficacies are uncertain.21 Decision support Monitor progression When the level of evidence, particularly of diagnostic and tools to Refine Predict/ Predict events help: Assess risk assessment diagnose Inform therapeutics therapeutic intervention effectiveness, improves significantly, real strides can be made in the overall quality of care. Evidence- Cost/irreversibility Baseline Initiating Earliest Earliest Typical current based approaches, which gather information from large risk events molecular clinical intervention Disease burden detection detection populations, can increase the likelihood of effectiveness, though they cannot guarantee the outcome. The same treatment can result in different outcomes among different patients due to difficulty in accurately diagnosing diseases based primarily on Time symptoms or due to different responses from patients to similar treatments. For example, one patient may respond well to a Baseline risk Preclinical Disease initiation and progression progression certain treatment, another may respond poorly, a third may Therapeutic have an adverse reaction and a fourth may not respond at all. Sources of new Stable genomics • Single nucleotide Dynamic genomics • Gene expression decision support biomarkers: polymorphisms • Proteomics • Haplotype • Metabolomics As health promotion and care delivery move into the realm of mapping • Clinical risk models • Gene sequencing • Molecular imaging PHC, they continue to evolve to more of a science than an art, with clinician decisions based on patient preferences and Personal health plan evidence gathered from subpopulations as similar to the patient Sources: Ralph Snyderman, M.D.; IBM Global Business Services; IBM as possible. As a result, PHC is more precise for diagnosis, and Institute for Business Value. treatment of disease is more predictable. With PHC, enough Figure 5: PHC has potential to provide better risk assessment patient information and clinical knowledge exist to diagnose a and prediction, earlier detection, earlier treatment and targeted disease – ideally as early as possible based on cause rather than treatments. on symptoms – and treatments exist to treat the causes of the
  12. 12. 10 IT-enabled personalized healthcare PHC with and without “-omics” Applying PHC to breast cancer Despite the strong potential of “-omics,” as illustrated by the Consider the case of breast cancer: Women with the breast cancer example described in the sidebar, a working BRCA1 or BRCA2 genes are at a significantly higher risk personalized healthcare system is by no means entirely for the disease.23 That information, gathered through ge- dependent on them for its success. However, including netic testing and combined with information about family “-omics” information and clinical knowledge can improve both history and personal lifestyle, provides a baseline for as- the individual clinician’s and the system’s effectiveness. We sessing a particular individual’s degree of risk. If the risk describe five ways to provide PHC by leveraging information is high, then the individual may require earlier or more fre- quent screenings or may be more willing to take preven- and knowledge available today and how each can be enhanced tive steps such as maintaining a healthy weight, exercis- with “-omics” information and knowledge: ing, eating well, limiting alcohol consumption and not smoking. 1) Incorporate patient preferences into decision making. Individuals can make their preferences known through tools The next step involves early detection. More information such as advance directives, which can include “Do Not about the molecular pathways through which tumor cells Resuscitate” agreements and instructions regarding life- circulate is being uncovered, with the hope that tests can be developed to diagnose the disease before traditional sustaining treatments such as cardiopulmonary resuscitation tests typically can. Meanwhile, clinical detection is con- (CPR), dialysis, artificial nutrition and mechanical ventilation. stantly improving, through the detection of lumps and Additionally, clinicians and patients can participate in shared mammography – possibly including magnetic resonance decision making. Physicians and other clinicians may increas- imaging (MRI) or positron emission tomography (PET) ingly be interpreters of information and facilitators of decisions scans for high-risk patients – and, increasingly, molecular rather than sole decision makers. In cases where there is no biomarkers (a characteristic that can be objectively mea- obvious best choice, clinicians can present information about sured and evaluated as an indicator of biologic process- benefits, risks and costs of treatment alternatives, and the es) that indicate disease onset prior to any phenotypic or patient can make decisions based on his or her personal values physical symptoms.24 Once the disease is detected, cur- and convey these decisions through tools such as informed rent therapies typically include surgery, radiation or che- consent. motherapy. Next, capabilities now exist to aid in therapeutic choices. Use of “-omics” will provide much more complete knowledge Overexpression of the HER2 gene in breast cancer pa- about the risks inherent in various diseases, as well as the tients, present in about 30 percent of patients, has been effectiveness of various treatments. This knowledge about associated with a poorer prognosis and a higher potential baseline risk and preventive or therapeutic intervention for recurrence of the cancer. The overexpression can be effectiveness will enable patients to collaborate even more suppressed by the use of the drug Herceptin® (trastuzum- ab).25 Similarly, by evaluating 21 separate gene expres- closely with clinicians to make vital care decisions with greater sions, the Oncotype DX® test can calculate the level of confidence. risk of the cancer returning within ten years and thus aid clinicians in making decisions regarding chemotherapy.26
  13. 13. IBM Global Business Services 11 2) Improve prediction and early detection. More complete information, including genetic information, Predictive modeling can help answer the question, “What will increase the need for tools to help identify and visualize might happen next?” Geisinger Health System uses predictive the relevant information in a way that aids clinical decision modeling for congestive heart failure (CHF) using nonge- making. The need for better clinical decision support systems nomic data contained in its EHR to identify patients who may for clinicians and expert team members such as geneticists will develop CHF. Roughly half of the patients were diagnosed 6 to also increase significantly as the volume and complexity of the 24 months before a diagnosis would typically have been data used for clinical decision making increase. determined, giving an opportunity to avoid the disease or lessen the impact.27 4) Develop multiple channels for delivering care. New knowledge about preventing, predicting, diagnosing, Data from patient monitoring equipment can also be useful in treating and managing diseases will likely expand the variety of predictive modeling. Analyzing such data can help identify delivery approaches and channels. Obviously, care delivery high-risk patients earlier, perhaps avoiding emergency visits or cannot be personalized unless patients are willing to receive even hospitalizations. such care, perhaps through more convenient and cost-effective venues than today’s ambulatory or inpatient facilities. Thus, it Employing ”-omics” techniques could enable more accurate will become more important to offer a wide range of care and earlier preclinical detection of a disease through delivery channels – from e-mail, e-visits and e-consultations to proteomics and complementary technologies such as molecular retail clinics, specialized treatment facilities, telemedicine and imaging. Added benefits could include reduced total cost for remote monitoring. detection and treatment since diseases could be treated at a less severe stage and improved potential to reverse diseases such as The increased use of “-omics” information and knowledge diabetes by detecting them at an earlier stage (see Figure 5). could help tailor channels based on better understanding of diseases and their causes, as well as effective prevention, 3) Support clinical decision making. detection and treatment options. For example, cancer EHRs can store large amounts of clinical information about treatment today is organized largely by body organ. In the individual patients. However, that information can be difficult future, treatment may be organized instead by molecular to locate, organize and use – particularly if it is outside the pathways, creating new delivery channels or treatment context of why it was originally captured. By using tools to centers.29 Also, as knowledge of diseases at the molecular level better integrate key data and then presenting them visually in continues to evolve, more treatment locations or channels, the appropriate decision-making context, clinicians can better such as outpatient care or home-based care, may become viable understand and use them. Geisinger found that physicians use for diseases that today must be treated in more expensive and 50 percent more data in making clinical decisions regarding intensive acute care settings. rheumatology when the information is organized and displayed properly.28
  14. 14. 12 IT-enabled personalized healthcare 5) Help activate and sustain lifestyle and behavior changes. Both prevention and disease management (DM) must be As chronic diseases become more prevalent, the role that cornerstones of any high-value healthcare system. While there lifestyle and behaviors play on health continues to grow in are many benefits associated with both, cost effectiveness is a importance – both in maintaining good health and managing a subject of considerable debate. We believe that more complete condition once it is diagnosed. Just as patients respond patient information and improved clinical knowledge could differently to treatments, they also are motivated by different greatly improve the cost effectiveness of prevention and disease approaches. To date, helping motivate individuals to make and management – and could even generate net cost savings. sustain lifestyle and behavior changes has not been a focus of many healthcare systems across the world and remains an area In hopes of preventing disease, for example, clinicians currently of great interest with a scarcity of knowledge. screen broad sections of the population to determine whether particular individuals have a specific disease. With PHC, Better information about genetic dispositions for disease and preventive activities such as selective screenings could be the genetic causes of disease can further this effort in multiple conducted based on clinical utility – in other words, with better ways. If an individual knows that he or she has genetic factors knowledge of risks and benefits – better matching preventive that increase risk for certain diseases, that knowledge could activities with individual risk profiles. help motivate lifestyle changes to prevent the disease. Addi- tionally, genetic factors may impact an individual’s ability to Similarly, today’s disease management efforts have not consis- make lifestyle changes. For example, early evidence indicates tently yielded the cost and quality benefits desired. Better that certain genes may make smoking cessation more information and knowledge, however, could enable specific difficult.30 Knowledge of these genetic factors could influence management approaches tailored to each individual or subpop- selection of approaches for helping the individual make and ulation of individuals, depending on their conditions, prefer- sustain changes. ences and goals to improve both the results and the cost effectiveness. For example, a clinician could more cost-effec- The value of IT-enabled PHC tively work with a patient to prevent him or her from PHC holds the potential to vastly improve the quality of becoming a Type 2 diabetic or work with a diabetic to better healthcare and the way it is delivered, while potentially manage the condition, rather than have a patient progress reducing its overall cost (see Figure 6). As previously stated, unimpeded to the point of needing kidney dialysis and foot PHC can help provide the right treatment to the right person amputations, ideally significantly reducing the total costs of at the right time through earlier and more precise diagnosis care over the duration of the condition. and cost-effective treatments – which may then improve patient compliance with treatment regimens. PHC has the potential to improve the cost effectiveness of many of today’s high-cost or low-benefit activities. PHC has the potential not only to improve the quality of healthcare, but also the overall costs and the cost-effectiveness of many of today’s high-cost or low-benefit activities.
  15. 15. IBM Global Business Services 13 With today’s information With IT-enabled PHC Diagnosis • 15 percent are inaccurate or incomplete • Better ability to distinguish among diseases with similar symptoms • 20 percent of fatal illnesses misdiagnosed • Better ability to diagnose based on cause rather than by symptoms • Ability to reinterpret patient data based on new clinical knowledge Treatment • Evidence for maybe one-third of what is done • Ability to generate and incorporate more and finer-grained evidence effectiveness • Where evidence exists, it is “one size fits all” • Tailored interventions • Lengthy delays to incorporate latest clinical • Better knowledge of when not to treat knowledge into practice • Better monitoring of patients and compliance • Poor patient compliance Prevention to • Screenings for broad populations • Selective screenings based on clinical utility avoid disease • Behavior changes hard to make and sustain • Stronger evidence of risk and better knowledge of what helps drive or compress sustainable behavior changes with similar people morbidity Disease • Difficult to identify which patients will benefit • Risks stratification based on clinical, environmental and genomic management most from different types of active management information to identify patients with maximum benefit Technology • New technology drives up costs (e.g., overuse • Evidence-based treatment plans, including diagnostics, required or misuse) Sources: IBM Global Business Services; IBM Institute for Business Value. Figure 6: PHC has potential to improve the quality and cost-effectiveness of many of today’s current activities – and possibly even to reduce costs. New technologies and treatments also present vexing cost and Exceeding human cognitive capacity quality problems. Healthcare is one of the few areas in which While better patient information and clinical knowledge offer new technologies often increase costs rather than reduce them. potential benefits, they also present a major obstacle to PHC. Cheaper diagnostic technologies, such as X-rays, may be The amount, complexity and diversity of information and replaced with much more expensive technologies, such as knowledge currently being generated will increase immeasur- computerized tomography (CT) scans, and cheaper therapies ably as research into “-omics” continues. Even today’s informa- may be replaced with more expensive therapies such as new tion and knowledge needed to make good clinical decisions are surgical approaches or new, expensive patented drugs. Yet, frequently well beyond the cognitive capacity of clinicians and knowledge is frequently missing or inaccessible to help patients. It has been estimated that the human mind typically practitioners understand whether or when the added benefit is can make use of no more than five to nine facts at a time when worth the additional cost. Better patient information and making decisions, such as about diagnoses or optimal treatment clinical knowledge could help determine comparative effective- regimens.31 Decisions involving patients with multiple chronic ness or cost effectiveness and then help develop and incorpo- conditions may require the ability to process as many as 100 or rate evidence-based diagnostic and treatment plans into care delivery processes.
  16. 16. 14 IT-enabled personalized healthcare more facts – and even more than 1,000 facts when information regarding proteomics and other disciplines at the molecular level are added to the equation. No clinician can function “Too often, U.S. healthcare overvalues local optimally without the aid of a sophisticated HIT environment autonomy and undervalues disciplined at this level of complexity (see Figure 7). science – not because of inattention or incompe- Through improved science, PHC has great potential to tence among doctors and nurses but because it improve quality and reduce costs of health promotion and care is difficult for the human mind to keep up delivery. But it is incredibly information and knowledge with the explosion of medical knowledge.” intensive – even compared to today’s already complex needs – The New England Journal of Medicine article 32 and exceeds human cognitive capacity. Access to and appro- priate use of burgeoning volumes of patient information and clinical knowledge will require a powerful HIT environment. The current state of HIT PHC will require a highly sophisticated HIT environment that can collect and analyze immense amounts of research and clinical information and knowledge and then present it in ways 1000 Proteomics and other that clinical decision makers can easily use. Unfortunately, the effector molecules current HIT environment is simply not up to such a task. It Facts per decision Functional genetics: currently addresses administrative needs (such as health plan 100 Gene expression profiles enrollment, physician or hospital billing and claims processing) Decisions for patients better than it does clinical needs. Simplistically, the current with multiple chronic conditions environment can be divided into four major sets of compo- 10 Structural genetics: nents or layers: care delivery systems, research systems, Human cognitive capacity e.g., SNPs, haplotypes 5 administrative systems and infrastructure (see Figure 8). Acute decisions by clinical phenotype 1990 2000 2010+ 2020+ Much of the hope for improving the current HIT environment rests on two types of digital patient records in the care delivery Sources: William Stead, MD; IBM Global Business Services; IBM Institute for layer: EHRs – the primary object of HITECH Act funding – Business Value. and personal health records (PHRs). Although these two terms Figure 7: A change in the nature of disease, plus an explosion are sometimes used interchangeably or even combined (the of clinical information and finer-grained clinical knowledge, electronic personal health record), we view them as distinct will challenge “expert- or experience-based practice.” entities.
  17. 17. IBM Global Business Services 15 Conceptual architecture of current healthcare IT environment Basic research Clinical trial Electronic data Research systems management capture Health information 38% Special-purpose Infrastructure exchanges networks 38% 38% 38% Care delivery Monolithic electronic Monolithic electronic Personal health health records health records records Administrative 38% Employer systems 38% Payer systems 38% Provider billing Sample interfaces shown only between layers Sources: IBM Global Business Services; IBM Institute for Business Value. Figure 8: The current HIT infrastructure does not lend itself to a learning, personalized, patient-centric healthcare system. While both are a “computer-accessible, interoperable resource and tools to support exchange of health information with other of pertinent health information on an individual,” an EHR is clinicians or with patients (for example, reminders or electronic used primarily by a broad spectrum of clinical personnel laboratory results). However, many of today’s EHRs are used involved in the individual’s care, enabling them to deliver and more to support coding of procedures and services and for coordinate care and promote the person’s wellness.33 A robust billing functions than to provide advanced clinical functions EHR system could include capabilities such as clinical docu- such as CPOE, clinical decision support or disease registries. mentation of problem, allergy and active medications lists; results viewing of laboratory, radiology or consultant reports; Similar to the EHR, the PHR contains pertinent health computerized provider order entry (CPOE) or e-prescribing; information on an individual.34 In contrast to the EHR, the clinical decision support such as clinical guidelines, reminders PHR is managed by the individual and is intended to supply and alerts; disease registries to help manage subpopulations of the information needed to help educate, empower and activate patients with similar diagnoses; portals to access the Internet; the individual to assume responsibility for his or her health and coordinate appropriately with health professionals.
  18. 18. 16 IT-enabled personalized healthcare The PHR may contain information that is not generally A variety of research systems are also in use, including systems included in EHRs today, such as the individual’s observations that manage clinical trials and electronic data capture systems of daily living or exercise logs. PHRs are still in their infancy designed to collect research data as part of a clinical study. today, with only 7 percent of respondents to a recent survey Increasingly, there is a focus on capturing data from clinical having one.35 Even so, a variety of organizations are offering trials directly from EHRs to avoid duplicate data entry, thereby PHR platforms or capabilities, including employers, health improving accuracy and efficiency. insurers (through payer-based health records), care delivery organizations (by allowing patients access to certain EHR The administrative layer is perhaps the most mature layer of information) and IT vendors such as Google Health and the HIT environment. It contains applications such as Microsoft. employer enrollment systems to support enrollment of employees into health plans, provider systems to support Somewhat simplistically, today’s infrastructure layer contains billing functions and payer systems to support claims health information exchanges (HIEs), also a target of processing and payments. HITECH funding, and special-purpose networks to support functions such as claims submission or payments. An HIE Even though these four layers of the U.S. HIT environment facilitates the electronic movement of any and all health- have been maturing over many years, adoption rates by related data according to an agreed-upon set of interoperability clinicians for EHRs, one of the most critical applications for standards, processes and activities across nonaffiliated organi- PHC, remain low – just 6 percent of hospitals, 20 percent of zations in a manner that protects the privacy and security of large physician groups and 8 percent of small groups have that data and of the entity that organizes and takes responsi- implemented advanced EHRs.38 Low adoption can be attrib- bility for the process.36 A 2009 survey identified 193 active uted to a number of factors, including the expense and tech- HIE initiatives in the United States, with 57 being operational. nology competency required to purchase, implement and Laboratory and medication data are the types most frequently maintain them; the impact on clinical workflow; and concerns exchanged.37 about interoperability. The HITECH Act is intended to help address the problems with low EHR adoption in the United States. PHR adoption could also remain low without EHRs and, possibly, HIEs available to help easily populate the Low adoption of EHRs can be attributed to a relevant PHR information. number of factors, including the expense and Even when EHR and PHR adoption improves significantly, technology competency required to purchase, the HIT environment will still face five interdependent implement and maintain them; the impact on challenges to support the vision and promise of PHC. clinical workflow; and concerns about interoperability.
  19. 19. IBM Global Business Services 17 Toward a new HIT environment Issues beyond the scope of this paper PHC holds the potential to reduce costs while increasing the While we believe that a patient-centric, high- quality and continued innovation of health systems. To achieve performance PHC system must be built on a strong HIT this goal, a variety of policy, education, funding and social environment, we recognize that many issues must be addressed concurrently but are beyond the scope of challenges must be addressed (see sidebar, Issues beyond the scope this paper. of this paper). At the same time, a much more open, robust, flexible standards-based HIT environment will be required. These issues include a variety of policy concerns, Additionally, not all of the technologies needed are readily such as privacy, informed consent, reimbursement, available today. New general-purpose and healthcare-specific ownership of intellectual property, licensing and technologies need to be developed.39 privileges for clinicians, standards for drug and device approval, and liability for using or not using evidence- This HIT environment must be capable of capturing, storing, based systems. Another set of issues involves analyzing and appropriately sharing information about educating stakeholders about PHC and the use of individual patients and patient populations. It also must be technologies that enable it. Clinicians and individuals capable of rapidly generating and managing new clinical will need to understand more about genomics, and knowledge and easily incorporating it into clinical processes caregivers must learn to accept computerized help and team-based care, for instance. Next, issues of how to and workflows for decision making related to health promotion fund these new systems and expectations regarding and care delivery. This environment also must facilitate the return on investment from them will be central to appropriate interactions among constituents. These capabilities successfully building them. And finally, there are serious were not top priorities when today’s HIT systems were ethical issues to face: Assuming such systems can be designed and implemented. Today’s systems were designed built, should they be? For example, what would be the primarily to facilitate administrative functions and to automate societal, financial and environmental impacts if virtually specific clinical encounters. everyone lived to be 110 years old? Five interdependent HIT challenges must be overcome for Also, a number of technology-related topics such as PHC to succeed: the future of hand-held devices, servers, networks and storage devices; policy issues such as information 1. Lack of an interoperable HIT environment for care delivery retention, security and compliance and approved secondary uses of data; and technical issues such as and research how best to manage and secure huge volumes of data 2. Prevalence of tightly coupled applications and data are best left to other more technical documents. 3. Inadequate data and knowledge standards 4. Insufficient analytics capabilities 5. Absence of a clinical decision-making foundation.
  20. 20. 18 IT-enabled personalized healthcare Challenge 1: Lack of an interoperable HIT tional organizational and industry boundaries. Next, post- environment for care delivery and market surveillance of patients undergoing new treatments and ongoing monitoring of the effectiveness of the treatments is research done only sporadically today and needs to become widespread. The HIT environment (see Figure 9) will have to be more This could require much tighter links between research flexible, functionally rich and interoperable (the ability of systems and EHRs or PHRs. different information technology systems and software applications to communicate; exchange data accurately, Infrastructure effectively and consistently; and use the information that has The infrastructure layer will have to become much more been exchanged) than today’s highly fragmented HIT environ- functionally rich and even perform some of the services that ment.40 Each layer – research, infrastructure, care delivery and historically might have been considered applications. administrative – will need changes. Several capabilities must be added or enhanced to facilitate a wider variety and volume of information exchange among Research stakeholders. The research layer will need a number of changes. First, basic research systems will need to facilitate research across tradi- Conceptual architecture of PHC-enabled healthcare IT environment Post-market Basic research Clinical trial Electronic data surveillance/38% Research 38% 38% 38% systems management capture effectiveness Specialized 38% Knowledge clouds 38% HIEs 38% 38% Patient monitoring Public health38% Infrastructure information sources* (e.g., genomics interpretations) 38% 38% 38% Care delivery EHR system EHR system PHRs Payer systems Administrative Employer systems 38% Provider billing 38% (PBHRs) 38% *Examples include genomics, biobanks and patient information. Sources: IBM Global Business Services; IBM Institute for Business Value. Figure 9: A more interoperable, flexible, functionally rich HIT environment will be needed to cross the information and knowledge chasm to PHC.
  21. 21. IBM Global Business Services 19 • Specialized information sources or information “clouds.” A wide from specialized information and knowledge sources, such as variety of critical data and information – including genomics PHRs or biobanks, among a larger group of entities and medical imaging, for instance – may be too sensitive, too (competing organizations, organizations outside the region or specialized or simply too large to incorporate directly into researchers, for example). EHRs. In this case, the information may need to be kept as a shared resource accessible through the Internet or a private Care delivery network. For the care delivery layer, EHRs will need different capabili- • Specialized knowledge sources or knowledge “clouds.” The ties, including access to specialized information or knowledge knowledge and skills required to interpret certain kinds of sources such as genomics information, specialized disease data, such as genomics or proteomics, will likely be specialized registries or biobanks and the ability to incorporate relevant enough that it will need to be shared across multiple EHRs, data from patient monitoring devices. facilities, organizations and care venues. • Patient monitoring. Realtime patient monitoring and ongoing EHRs also will need additional functionality and interoper- surveillance – regardless of location (in a hospital or at home, ability capabilities to support a broader scope of research for example) – will be necessary to track current health efforts. EHRs can be beneficial for efficacy research and are conditions, adherence to treatment programs and longer-term essential for effectiveness research by tracking clinical outcomes. outcomes over time, thereby more closely integrating care • Public health. Systems will be required to track the overall delivery and research (see sidebar, Efficacy vs. effectiveness quality and safety of treatments, as well as to manage public research). Comparing effectiveness of alternative interventions health reporting, such as disease outbreaks and the results of or approaches to health promotion and care delivery – in other large-scale efforts to promote health and prevent disease. words, comparative effectiveness research – may be a tool for • Health information exchanges (HIEs). Today’s HIEs provide intelligent cost containment and identifying preferred useful functions, such as exchanging lab or medication data therapies.42 and information about outpatient episodes or inpatient visits, frequently on a local or regional basis among organizations These EHR research and care delivery capabilities will likely with trusted relationships.41 As with other components, future require a very different technology architecture. Today’s HIEs may look quite different. They will likely need to “monolithic” EHRs can be difficult to change or to link with facilitate the sharing of a much wider variety of information other systems within or external to the organization (depart- mental systems in a hospital or physician office systems, for example). To more fully support PHC and other requirements, future EHRs likely will have to be composed of a number of components or services that can work seamlessly together and EHRs will need additional functionality and can more easily be changed. EHR vendors likely will encounter interoperability capabilities and will likely major obstacles in “rearchitecting” EHRs from a monolithic to require a very different technology a more service-oriented architecture. These efforts could take architecture. over ten years to complete.43
  22. 22. 20 IT-enabled personalized healthcare Today’s nascent PHRs will need to become much more mature, Efficacy vs. effectiveness research functionally rich and interoperable, becoming the individual’s Efficacy research focuses on the extent to which window or portal to the healthcare world in a more patient- a health care intervention is beneficial over the centric environment. Possible uses for the consumer-controlled short term, often compared to a placebo, when PHR include interacting with clinicians (for example, administered in an idealized setting to a small group conducting e-visits or scheduling appointments) or genetics of carefully selected, highly compliant patients. counselors; capturing and analyzing information from remote Phase II and III clinical trials often focus on efficacy. monitoring devices and sharing relevant information with Effectiveness research focuses on the extent to which provider EHRs; helping monitor compliance to treatment a health care intervention works, possibly compared regimens; connecting with patients who have similar condi- to other viable interventions, over the longer term tions; capturing self-reported data such as observations of daily living; and linking to additional knowledge about diseases, when provided to a wide assortment of real-world disease prevention or general health promotion. patients, including those with multiple conditions, in diverse clinical settings across the country. Administrative Of course, the administrative systems must change to support this new environment. For example, as more information about the causes and treatments of diseases becomes available, payer/ insurance systems and provider billing systems will need to support new ways of classifying diseases and new therapies or combinations of diagnostics and therapies. Also, for the near future, payer/insurance systems may have a broader range of information about individual patients than today’s fragmented EHR systems. For example, payer systems may have informa- tion regarding medication prescriptions, visits to other care providers or potential gaps in care. This type of information needs to be shared with the EHR and PHR systems on a wider, more consistent basis than it is today. Together, these changes to the HIT environment will help lay the foundation for a rapid-learning environment with better integration across care delivery and research.
  23. 23. IBM Global Business Services 21 Challenge 2: Prevalence of tightly coupled Data Data Clinical Data applications and data acquisition integration intelligence access and decision The new architecture previously described depends in large support part on the ability of a wide range of applications and tools to External data Integrated Clinical Researchers gain access to necessary data. Somewhat simplistically, an Medline, information intelligence genomics, infrastructure Evidence application consists of three major layers. The presentation disease Security, generation, layer controls data input and output to the application user data, clinical anonymization, data mining guidelines, auditing, and through technologies such as a PC screen, Internet browser clinical trials, privacy, collaboration adverse drug federation, or mobile device. The application layer contains the business reaction, etc. warehousing, Administrators Guideline or clinical logic (what laboratory tests need to run, for curation, management ontologies, example) and work flow (who needs to approve or process a data models Evidence Internal data and rules laboratory order, for example). The data layer contains the EMR, management data used by the application. ambulatory, lab, pharmacy, imaging, Decision Physicians financial, In the current environment, applications and the data they administrative, support Dx, Rx Rules-based depend on are typically very closely intertwined. The data etc. report and alert generation Patients generated by EHRs, for instance, often are not easily available to caregivers who don’t have access to that EHR, to the patient Access and collaborative or to researchers outside the care delivery organization who Patients could benefit from appropriate access. Building tools to extract data from the originating application can be a laborious, costly Sources: IBM Global Business Services; IBM Institute for Business Value. and painful process. Only by decoupling or “liberating” such data from the applications that generate or create them can Figure 10: Data and applications must be decoupled for robust use of they be used throughout the HIT environment for care the data and for applications to draw upon multiple data sources. delivery or research – with appropriate data integrity, privacy and security capabilities in place (see Figure 10). • Data for individual patients (including images and tissue samples) could be compared with data for a large number of Making relevant data within the system available to key other patients to help diagnose or determine the best stakeholders would result in a number of benefits vital to a treatment options. successful PHC system: • Doctors and researchers could collaborate across organizational boundaries applying data mining and analysis • Relevant information for a specific patient, regardless of when to the entire set of clinical patient information to develop or where it was generated, could be combined to provide a disease-specific, personalized clinical decision rules for more complete picture of the patient’s health. diagnosing and treating patients.